"The 'data' object was created by an older version of PyG. "

[Neo-Flash]

Hi!
thank you for sharing your work
When I ues google colab
run"python -u script/gen.py --config_path ./config/qm9_default.yml --generator ConfGF --smiles c1ccccc1"
it returned an error :RuntimeError: The 'data' object was created by an older version of PyG. If this error occurred while loading an already existing dataset, remove the 'processed/' directory in the dataset's root folder and try again.

could you please tell me how to fix this?

whole output:

Let's use 1 GPUs!
Using device cuda:0 as main device
{'train': {'batch_size': 128, 'seed': 2021, 'epochs': 300, 'shuffle': True, 'resume_train': False, 'eval': True, 'num_workers': 0, 'gpus': [0], 'anneal_power': 2.0, 'save': True, 'save_path': '/home/shichenc/scratch/confgf/train', 'resume_checkpoint': None, 'resume_epoch': None, 'log_interval': 400, 'optimizer': {'type': 'Adam', 'lr': 0.001, 'weight_decay': 0.0, 'dropout': 0.0}, 'scheduler': {'type': 'plateau', 'factor': 0.6, 'patience': 10, 'min_lr': '1e-4'}, 'device': device(type='cuda', index=0)}, 'test': {'init_checkpoint': '/home/shichenc/scratch/confgf/train/qm9_default', 'output_path': '/home/shichenc/scratch/confgf/test/qm9_default', 'epoch': 284, 'gen': {'dg_step_size': 3.0, 'dg_num_steps': 1000, 'steps_d': 100, 'step_lr_d': 2e-06, 'steps_pos': 100, 'step_lr_pos': 2.4e-06, 'clip': 1000, 'min_sigma': 0.0, 'verbose': 1}}, 'data': {'base_path': '/content/', 'dataset': 'qm9', 'train_set': 'train_data_40k.pkl', 'val_set': 'val_data_5k.pkl', 'test_set': 'test_data_200.pkl'}, 'model': {'name': 'qm9_default', 'hidden_dim': 256, 'num_convs': 4, 'sigma_begin': 10, 'sigma_end': 0.01, 'num_noise_level': 50, 'order': 3, 'mlp_act': 'relu', 'gnn_act': 'relu', 'cutoff': 10.0, 'short_cut': True, 'concat_hidden': False, 'noise_type': 'symmetry', 'edge_encoder': 'mlp'}}
set seed for random, numpy and torch
loading data from /content/qm9_processed
train size : 0 || val size: 0 || test size: 24068
loading data done!
got 200 molecules with 24068 confs
Traceback (most recent call last):
File "script/gen.py", line 92, in
test_data = dataset.GEOMDataset_PackedConf(data=test_data, transform=transform)
File "/content/ConfGF/confgf/dataset/dataset.py", line 449, in init
self._pack_data_by_mol()
File "/content/ConfGF/confgf/dataset/dataset.py", line 469, in _pack_data_by_mol
data = copy.deepcopy(v[0])
File "/usr/local/lib/python3.7/copy.py", line 161, in deepcopy
y = copier(memo)
File "/usr/local/lib/python3.7/site-packages/torch_geometric/data/data.py", line 392, in deepcopy
out._store._parent = out
File "/usr/local/lib/python3.7/site-packages/torch_geometric/data/data.py", line 358, in getattr
"The 'data' object was created by an older version of PyG. "
RuntimeError: The 'data' object was created by an older version of PyG. If this error occurred while loading an already existing dataset, remove the 'processed/' directory in the dataset's root folder and try again.

[Neo-Flash]

https://colab.research.google.com/drive/12L-XYWhcbrIKJEMNPOOkmLDb2DtGgn9t?usp=sharing

and this is my colab notebook ↑

[shichence]

Thank you for sharing your solutions. The ConfGF was implemented with the old-version Pyg. We plan to integrate the ConfGF into our TorchDrug platform in the near future.